Grant: Canada Research Chairs

Intelligent software systems (ISS), which are systems that integrate AI, are increasingly being deployed in large-scale and critical systems. Ensuring the trustworthiness of ISS is critical as evidenced by many recent deadly incidents such as an Uber self-driving car that ran into a pedestrian even though the car’s sensors detected her presence. To build trust in ISS, we should ensure that the decisions made by its components (including AI-powered components) are reliable, fair, and cannot cause harm. This Chair aims at developing theories, methods, and tools to support the development, deployment, maintenance, and evolution of trustworthy ISS. The following five research objectives (RO) are planned: (RO1) Identification of quantifiable attributes of trustworthiness—this first objective is important to bridge the gap between high-level principles of trustworthiness and their concrete manifestation in ISS. (RO2) Integration of trustworthiness requirements in the architecture and design of ISS—we will develop a validated catalog of architectural and design patterns for designing trustworthy ISS, and create a taxonomy mapping trustworthiness requirements to concrete architecture and design tactics, to guide ISS development teams. (RO3) Development of methodologies and tools to support the verification and validation of trustworthiness properties throughout the lifecycle of ISS—we will devise software testing mechanisms that can verify trustworthiness properties such as correctness, robustness, security, privacy, and fairness/bias of ISS. (RO4) Development of methodologies and tools to support trustworthiness operations in production— we will develop methodologies and techniques for building automated pipelines to deploy, operate, monitor, and continuously improve ISS running in different deployment environments (e.g., centralized servers and edge devices). (RO5) Development of methodologies and tools to ensure the trustworthiness of AI models supply chains—we will examine the decision-making workflow and challenges of AI model reuse from third-party model registries and develop techniques and tools to support the safe and trustworthy reuse of AI models. The results of the Chair will support Canadian organizations in developing trustworthy intelligent systems that will help improve the quality of life of Canadians. It will also provide them with a competitive advantage in the booming AI market.